MDR for improved fluoroelastomer cure testing.

Minnesota Mining and Manufacturing (3M0 selected Monsanto's MDR2000 moving die rheometer for cure testing of Fluorel fluoroelastomers. The selection of the MDR2000 resulted from its reduced testing variability and improved discrimination of compound differences in fluoroelastomers. The improved performance and productivity of this rheometer have justified the replacement of existing rheometers with the MDR2000.

3M evaluated an MDR2000 as compared to Monsanto's R100S type oscillating disc rheometers and an ODR2000. The MDR2000 related twice as high as either the R100S or ODR2000 for discriminating compound differences. The variability of the MDR2000 was typically 2-4 times less than the other rheometers.

Two studies in 1991 compared the instruments first at the 3M specialty fluoropolymers laboratory and then at two additional 3M locations. All locations had access to each instrument type, so that location effects could be separated from machine effects.

Variations of a Fluorel compound were made with 10% change in Curative A and Curative B. The experiment used a factorial design with the standard Fluorel FC-2174 gum, 30 parts of N990 carbon black, 3 parts of MgO and 6 parts of Ca(OH)2.

Starting with standard FC-2174 fluoroelastomer gum, levels of Curative A and Curative B were varied in a two level factorial design to help determine the discriminating power of each Machine-Arc combination. Table 1 shows the level number associated with the Curative A, Curative B combinations.

Table 1 - curative levels
Level  Curative A   Curative B
  1       -10%          10%
  2         0%           0%
  3        10%          10%
  4       -10%         -10%
  5        10%         -10%

Each instrument was calibrated prior to testing. All rheometer tests were run at 177 [degrees] C for 12 minutes using one mil polyester film. The test sample sizes were eight grams for the MDR2000 and 15 grams for the R100S and ODR2000.

The measured rheometer properties for FKM were minimum torque, maximum torque, ts2, t'50 and t'90. Tables 2 through 6 show the discriminating power (sensitivity) for each property. The variation for each machine was determined by pooling each location into one overall estimate. Evidence of unequal variation among locations for the same machine are noted at the bottom of each table.

Table 2 - minimum torque discrimination table
Machine               Average     Pooled    Ratio
                        range    std. dev.
MDR-2000, arc - 0.5  0.148333      0.010  14.833
ODR-2000, arc = 1.0   0.74357   0.073416  10.128(*)
ODR-100S, arc = 3.0   1.63333     0.1253  13.035(*)
(*) - Evidence of outliers (problem points) (location 2)
+ - Evidence of unequal variance between locations
    (location 2)
Table 3 - maximum torque discrimination table
Machine             Average      Pooled     Ratio
                      range     std. dev.
MDR-2000, arc = 0.5   3.875     0.1253     30.926
ODR-2000, arc = 1.0 7.38167   0.406202    18.172+
ODR-100S, arc = 3.0   15.75   0.806226    19.535+
+ - Evidence of unequal variance between locations
    (location 2, location 3)
Table 4 - ts2 discrimination table
Machine               Average      Pooled     Ratio
                        range     std. dev.
MDR-2000, arc = 0.5   0.31086   0.0106771    29.115
ODR-2000, arc = 1.0   0.45666      0.0402   11.36(++)
ODR-100S, arc = 3.0  0.466667      0.0683    6.8326
++ - Marginal evidence of unequal variance between
    locations (location 2)
Table 5 - t'50 discrimination table
Machine               Average       Pooled      Ratio
                        range    std. dev.
MDR-2000, arc = 0.5   0.38741   0.00844985      45.85
ODR-2000, arc = 1.0  0.608333     0.053386   11.395(++)
ODR-100S, arc = 3.0     0.650     0.065803      9.878
++ - Marginal evidence of unequal variance between
    locations (location 2)
Table 5 - t'90 discrimination table
Machine              Average   Pooled       Ratio
                      range    std. dev.
MDR-2000, arc = 0.5  0.57293   0.0211    27.153(+)
ODR-2000, arc = 1.0  0.65333   0.0543   12.032(++)
ODR-100S, arc = 3.0  0.66667   0.0585     11.396
+ - Evidence of unequal variance between locations
    (location 2)
++ - Marginal evidence of unequal variance between
    locations (location 2)

[Tabular Data Omitted]

The range of values, especially for minimum and maximum torques, change with instrument type. To compare instrument types for the ability to discriminate compound differences, the range for each type needs to be considered.

Taking a ratio of the range across levels (maximum minus minimum) divided by the testing pooled standard deviation is a good way to measure the ability to discriminate. This ratio gives the number of test standard deviations that fit between the maximum and minimum response levels for each change in curative level.

Table 7 (discriminate grading grid) grades each instrument type by rheometer data point. For each data point, the best discrimination (largest ration number) is a grade of 3, and the next best a 2. The sum across data point grades gives a "Sum" for each instrument and oscillation arc. This table shows the MDR2000 better discriminates between the cure systems and had less variability than the R100S or ODR2000 rheometers.

The method of ranking gives equal weight to each property and not accounting for ratios that are close seconds within a property. Asterisks were placed where potential outliers existed. One could also adjust the values base on potential outliers, asterisk values. Also no potential outliers were indicated with any of the MDR or ODR2000 series.